National Student Clearinghouse - Enrollment

Data Prep

Okay, lets check out the National Student Clearinghouse enrollment data. I will first import the entire dataset, clean it up, and then merge it with the master / Southern graduate dataset.

nsc.enrollment.original <- read_csv('/Users/kellyasche/Library/CloudStorage/GoogleDrive-kasche@ruralmn.org/My Drive/Research/FY25-27/FY26/OHE Journey to meaningful employment/Data/SLEDS/NSCEnrollment/NSC_Enrollments.csv') %>%
  mutate(PersonID = as.integer(PersonID),
         OPEID = str_pad(OPEID, width = 8, side = "left", pad = "0"),
         OPEID.6 = str_sub(OPEID, 1, 6)) %>%
  drop_na(PersonID) %>%
  drop_na(OPEID)
  
kable(head(nsc.enrollment.original))

PersonID	EnrollmentBeginTimeID	EnrollmentEndTimeID	OPEID	EnrollmentStatus	OPEID.6
10025234	20120103	20120126	00235400	H	002354
11364693	20130122	20130517	00238881	F	002388
3090532	20090824	20091218	00238300	F	002383
11127041	20200601	20200819	00238200	L	002382
4465956	20100823	20101217	00299700	F	002997
9191392	20110110	20110513	00553400	L	005534

kable(names(nsc.enrollment.original))

x
PersonID
EnrollmentBeginTimeID
EnrollmentEndTimeID
OPEID
EnrollmentStatus
OPEID.6

The dataset has 3,010,265 rows and 6 columns. Here are the columns and their definitions.

• EnrollmentBeginTimeID: Begin date for the student’s period of attendance.
• EnrollmentEndTimeID: End date for the student’s period of attendance.
• OPEID: Office of Postsecondary Education (OPE)/FICE code of the college that the student attended (Foreign Key to IPEDSCharacteristics).
• InstitutionName: Name of institution
• EnrollmentStatus: The last enrollment status reported for the student. This field will have ‘N/A’ or “NULL” if the reporting college has not defined the student’s enrollment status as directory information. Here are the code definitions;
  • F: Full-time
  • Q: Three-quarter time
  • H: Half-time
  • L: Less than half-time
  • A: Leave of absence
  • W: Withdrawn
  • D: Deceased

Essentially, this dataset provides semester-based information - each observation is a semester/single period with the institutional information and beginning and end time of that single period (usually a semester). For example, if an individual attended South Dakota State University, attended in a “typical” fashion (fall and spring semesters), and graduated in 4 years, there would be 8 observations for that PersonID - one observation per semester.

There are a few pieces of information that I want to consolidate from this dataset;

1. Did the PersonID in the master dataset attend a college - yes or no?
2. Did the PersonID attend college during high school (PSEO), immediately after graduation or wait?
3. Did the PersonID attend multiple colleges - yes or no?
4. What type of college did the PersonID attend first?
5. Did the PersonID attend a public, private not-for-profit, or private or, for-profit college(s)?
6. At any point during their college career, did the PersonID attend a college(s) inside or outside of the Southwest Region or outside the EDR of their high school?

I was going to include the length of time that a PersonID attended college, but due to differences in how institutions report enrollment periods, it didn’t seem like it would be a great indicator.

The pieces of information missing in the original dataset is location and type of institution attended.To do this we will need to join the original dataset with an IPEDS dataset using opeid.

The IPEDS data and the NSC data will only match using the first 6 digits of the IPEDS OPEID values. So I will change that in the NSC data. Then I can join.

ipeds.original <- read_csv('/Users/kellyasche/Library/CloudStorage/GoogleDrive-kasche@ruralmn.org/My Drive/Research/FY25-27/FY26/OHE Journey to meaningful employment/Data/SLEDS/IPEDS/IPEDS.csv') 

ipeds.sector <- ipeds.original %>%
  filter(!is.na(OPEID)) %>%
  mutate(OPEID = str_pad(OPEID, width = 8, side = "left", pad = "0"),
         OPEID.6 = str_sub(OPEID, 1, 6)) %>%
  select(Unitid, OPEID.6, City, State, FIPS, GeographicRegion, CountyCode, InstitutionSector, InstitutionName)  %>%
  mutate(CountyCode = str_pad(CountyCode, width = 5, side = "left", pad = "0"))

nsc.enrollment.ipeds<- nsc.enrollment.original %>%
  left_join(ipeds.sector, by = "OPEID.6") %>%
  drop_na(Unitid) %>%
  mutate(EnrollmentStatus = as_factor(EnrollmentStatus))

kable(head(nsc.enrollment.ipeds))

PersonID	EnrollmentBeginTimeID	EnrollmentEndTimeID	OPEID	EnrollmentStatus	OPEID.6	Unitid	City	State	FIPS	GeographicRegion	CountyCode	InstitutionSector	InstitutionName
10025234	20120103	20120126	00235400	H	002354	173665	Saint Paul	MN	27	4	27123	2	Hamline University
11364693	20130122	20130517	00238881	F	002388	174233	Duluth	MN	27	4	27137	1	University of Minnesota-Duluth
3090532	20090824	20091218	00238300	F	002383	174862	Saint Bonifacius	MN	27	4	27019	2	Crown College
11127041	20200601	20200819	00238200	L	002382	174844	Northfield	MN	27	4	27131	2	St Olaf College
4465956	20100823	20101217	00299700	F	002997	200332	Fargo	ND	38	4	38017	1	North Dakota State University-Main Campus
9191392	20110110	20110513	00553400	L	005534	174756	Saint Cloud	MN	27	4	27145	4	St Cloud Technical and Community College

kable(names(nsc.enrollment.ipeds))

x
PersonID
EnrollmentBeginTimeID
EnrollmentEndTimeID
OPEID
EnrollmentStatus
OPEID.6
Unitid
City
State
FIPS
GeographicRegion
CountyCode
InstitutionSector
InstitutionName

After joining these we now have 3,009,843 rows and 14 columns. This is a few less than the original enrollment document due to a few OPEIDs not aligning. But overall, the data now has the institution sector that the individual attended as well as location.

First, we will create a dataset with a column confirming their post-secondary attendance for each unique PersonID in the nsc.enrollment.ipeds.

nsc.enrollment.confirm <- nsc.enrollment.ipeds %>%
  filter(!EnrollmentStatus %in% c("A", "W", "D")) %>%
  distinct(PersonID) %>%
  mutate(attended.ps = "Yes")

kable(head(nsc.enrollment.confirm))

PersonID	attended.ps
10025234	Yes
11364693	Yes
3090532	Yes
11127041	Yes
4465956	Yes
9191392	Yes

kable(names(nsc.enrollment.confirm))

x
PersonID
attended.ps

There are 388,959 rows and 2 columns in the dataset. The columns provide the unique PersonID along with a newly created column confirming that they attended post-secondary education.

Next we will extract the first month and year each PersonID attended a post-secondary institution which will help us figure out how long after high school graduation did they wait until they attended a post-secondary institution.

nsc.enrollment.first.attend <- nsc.enrollment.ipeds %>%
  mutate(first.year = as.integer(str_sub(EnrollmentBeginTimeID, 1,4)),
         first.month = as.integer(str_sub(EnrollmentBeginTimeID, 5,6)),
         first.day = as.integer(str_sub(EnrollmentBeginTimeID, 7, 8))) %>%
  group_by(PersonID) %>%
  filter(first.year == min(first.year)) %>%
  mutate(n = n()) %>%
  ungroup() %>%
  group_by(PersonID) %>%
  filter(first.month == min(first.month)) %>%
  ungroup() %>%
  group_by(PersonID) %>%
  filter(first.day == min(first.day)) %>%
  ungroup() %>%
  distinct(PersonID, .keep_all = TRUE) %>%
  mutate(first.month = str_pad(first.month, width = 2, side = "left", pad = "0"),
         first.day = str_pad(first.day, width = 2, side = "left", pad = "0"),
         first.attend.ps = paste(first.year, first.month, first.day, sep = "")) %>%
  select(PersonID, first.attend.ps)

kable(head(nsc.enrollment.first.attend))

PersonID	first.attend.ps
3090532	20090824
6790768	20080825
9172836	20060906
1597967	20060905
5583982	20100823
5456903	20230109

kable(names(nsc.enrollment.first.attend))

x
PersonID
first.attend.ps

This dataset has 397,372 rows and 2 columns. Essentially, this dataset provides each unique PersonID with the earliest begin time for post-secondary education.

Next we will determine how many different institiutions the PersonID attended during their post-secondary career.

nsc.enrollment.n.institutions <- nsc.enrollment.ipeds %>%
  group_by(PersonID) %>%
  distinct(OPEID.6, .keep_all = TRUE) %>%
  ungroup() %>%
  group_by(PersonID) %>%
  summarize(n.institutions = n()) %>%
  ungroup()

kable(head(nsc.enrollment.n.institutions))

PersonID	n.institutions
14	1
98	1
141	1
198	2
208	1
227	2

kable(names(nsc.enrollment.n.institutions))

x
PersonID
n.institutions

As expected, we have 397,372 rows and 2 columns. Each PersonID in the dataset has the number of unique post-secondary institutions they attended.

Next, we will create a dataset indicating the type of college the PersonID attended first. I will use the IPEDS sector data. Here are the definitions of the institution sector;

• 0 - Administrative Unit
  
• 1 - Public, 4-year or above
  
• 2 - Private not-for-profit, 4-year or above
  
• 3 - Private for-profit, 4-year or above
  
• 4 - Public, 2-year
  
• 5 - Private not-for-profit, 2-year
  
• 6 - Private for-profit, 2-year
  
• 7 - Public, less-than 2-year
  
• 8 - Private not-for-profit, less-than 2-year
  
• 9 - Private for-profit, less-than 2-year
  
• 99 - Sector unknown (not active)

nsc.enrollment.first.institution <- nsc.enrollment.ipeds %>%
  mutate(EnrollmentBeginTimeID = ymd(EnrollmentBeginTimeID)) %>%
  group_by(PersonID) %>%
  filter(EnrollmentBeginTimeID == min(EnrollmentBeginTimeID)) %>%
  ungroup() %>%
  distinct(PersonID, .keep_all = TRUE) %>%
  select(PersonID, InstitutionSector) %>%
  rename(first.InstitutionSector = InstitutionSector)

kable(head(nsc.enrollment.first.institution))

PersonID	first.InstitutionSector
3090532	2
6790768	4
9172836	2
1597967	1
5583982	4
5456903	4

kable(names(nsc.enrollment.first.institution))

x
PersonID
first.InstitutionSector

As expected, we have 397,372 rows and 2 columns. This dataset now provides the sector of the first post-secondary institution they attended immediately after college.

Next we will determine thy type(s) of college(s) they attended during their post-secondary career. To do this we will first create a dataset with columns for each institution sector and a confirmation indicator on whether the PersonID attended that particular sector at some point in their career. I will then create a new category indicating that attended more than one type of institution sector. Once completed, I will be left with a dataset that has a column for each unique PersonID and what sector they attended, as well as a newly created code for “attended more than 1 type of sector”.

grad.list <- read_csv("Data/SLEDS/Masters/Master-8.csv") %>%
  select(PersonID)

nsc.enrollment.institution.types <- nsc.enrollment.ipeds %>%
  distinct(PersonID, InstitutionSector) %>%
  mutate(attended = 1) %>%
  complete(PersonID, InstitutionSector, fill = list(attended = 0)) %>%
  mutate(label = paste("ps.sector.", InstitutionSector, sep = "")) %>%
  select(PersonID, label, attended) %>%
  spread(label, attended) %>%
  right_join(grad.list, by = "PersonID") %>%
  gather(key = "key", value = "value", 2:10) %>%
  mutate(value = ifelse(is.na(value), 0, value)) %>%
  spread(key, value)

kable(head(nsc.enrollment.institution.types))

PersonID	ps.sector.1	ps.sector.2	ps.sector.3	ps.sector.4	ps.sector.5	ps.sector.6	ps.sector.7	ps.sector.8	ps.sector.9
55	0	0	0	0	0	0	0	0	0
181	0	0	0	0	0	0	0	0	0
280	1	0	0	0	0	0	0	0	0
301	1	0	0	0	0	0	0	0	0
304	0	0	0	0	0	0	0	0	0
412	1	0	0	0	0	0	0	0	0

kable(names(nsc.enrollment.institution.types))

x
PersonID
ps.sector.1
ps.sector.2
ps.sector.3
ps.sector.4
ps.sector.5
ps.sector.6
ps.sector.7
ps.sector.8
ps.sector.9

As expected, we have 122,938 rows and 10 columns. This dataset provides each unique PersonID with the institution sector they attended using the codes listed above. For PersonID’s that attended multiple sectors, they were coded as “10”.

Next we will determine whether they attended a post-secondary institution inside or outside of the planning region, outside their EDR, or outside of Minnesota. Since many of the PersonID in the dataset have attended multiple institutions, we will categorize it in the following way in order to capture the combinations of attendance;

• Attended inside region only (planning region, EDR, RUCA, State)
• Attended inside and outside region (planning region, EDR, RUCA, state)
• Attended outside region only (planning region, EDR, RUCA, state)

In order to do this we will need to combine our planning region and EDR joining documents with the nsc.enrollment.ipeds dataset. I need to remember that the way I arranged the regions are going to be different than the official planning region. We will also need to join the master dataset with it to determine the location of the PersonID’s high school graduation location. Lastly, we will need to join up the RUCA categories for counties outside of Minnesota. Then we can start the categorization process.

dem.desc.county <- read_csv("Data/Join docs/Master-ruca-county.csv")

grad.location <- read_csv("Data/SLEDS/Masters/Master-8.csv") %>%
  select(PersonID, edr, Dem_Desc) %>%
  rename(grad.edr = 2,
         grad.ruca = 3) %>%
  mutate(grad.pr = "Southern")

nsc.enrollment.location.join <- nsc.enrollment.ipeds %>%
  select(PersonID, InstitutionName, CountyCode) %>%
  left_join(dem.desc.county, by = c("CountyCode" = "State-County FIPS Code")) %>%
  left_join(grad.location, by = "PersonID") %>%
  drop_na(grad.ruca) %>%
  mutate(countyfp = str_sub(CountyCode, 3, 5),
         statefp = str_sub(CountyCode, 1, 2)) %>%
  left_join(counties.regions.1[,c(1,5,10,9)], by = c("countyfp", "statefp")) %>%
  rename(ps.Dem_Desc = 4,
         ps.countyfp = 8,
         ps.statefp = 9,
         ps.edr = 10,
         ps.pr = 11) 

kable(head(nsc.enrollment.location.join))

PersonID	InstitutionName	CountyCode	ps.Dem_Desc	grad.edr	grad.ruca	grad.pr	ps.countyfp	ps.statefp	ps.edr	ps.pr
10025234	Hamline University	27123	Urban/town/rural mix	EDR 10 - Southeast	Entirely urban	Southern	123	27	EDR 11 - 7 County Twin Cities	Seven County Mpls-St Paul
3090532	Crown College	27019	Entirely urban	EDR 10 - Southeast	Entirely urban	Southern	019	27	EDR 11 - 7 County Twin Cities	Seven County Mpls-St Paul
5943119	University of Minnesota-Twin Cities	27053	Entirely urban	EDR 10 - Southeast	Urban/town/rural mix	Southern	053	27	EDR 11 - 7 County Twin Cities	Seven County Mpls-St Paul
7693352	Azusa Pacific University	06037	Urban/town/rural mix	EDR 10 - Southeast	Town/rural mix	Southern	037	06	NA	NA
289664	Northwestern College	19167	Town/rural mix	EDR 9 - South Central	Urban/town/rural mix	Southern	167	19	NA	NA
9116706	Southwest Minnesota State University	27083	Town/rural mix	EDR 10 - Southeast	Urban/town/rural mix	Southern	083	27	EDR 8 - Southwest	Southwest

kable(names(nsc.enrollment.location.join))

x
PersonID
InstitutionName
CountyCode
ps.Dem_Desc
grad.edr
grad.ruca
grad.pr
ps.countyfp
ps.statefp
ps.edr
ps.pr

This joined dataset gives us 694,684 rows and 11 columns. The columns beginning with “ps” are the ruca category and regions of the post-secondary institution attended. The columns beginning with “grad” are the ruca category and regions of the high school from which the PersonID graduated.

One thing that’s important is to realize that joining the nsc.enrollment dataset with the Southern graduates dataset means we will have some NAs. Some of the students listed in the nsc.enrollment dataset aren’t in the Southern graduates dataset since they may have graduated before 2008 or didn’t actually meet the criteria in the Southern graduates dataset. Therefore, as we move forward, I will need to make sure to remove those NAs.

From here we can start creating new columns beginning with the RUCA category. to create this column we will gather each PersonID to examine whether or they attended a post-secondary institution in the same RUCA category, or if they attended multiple post-secondary institutions with one institution in the same category and another not the same.

nsc.enrollment.same.ruca <- nsc.enrollment.location.join %>% 
  mutate(ps.in.same.ruca = ifelse(ps.Dem_Desc == grad.ruca, "In same RUCA", "Outside RUCA")) %>%
  select(PersonID, ps.Dem_Desc, grad.ruca, ps.in.same.ruca) %>%
  select(PersonID, ps.in.same.ruca) %>%
  distinct(PersonID, ps.in.same.ruca) %>%
  mutate(code = ifelse(ps.in.same.ruca == "In same RUCA", 1, 2)) %>%
  group_by(PersonID) %>%
  summarize(code = sum(code)) %>%
  ungroup() %>%
  mutate(ps.in.same.ruca = ifelse(code == 1, "In same RUCA",
                                  ifelse(code == 2, "Outside RUCA", "Inside and outside same RUCA"))) %>%
  select(PersonID, ps.in.same.ruca)

kable(head(nsc.enrollment.same.ruca))

PersonID	ps.in.same.ruca
280	In same RUCA
301	Outside RUCA
412	Outside RUCA
513	Outside RUCA
687	Outside RUCA
710	In same RUCA

kable(names(nsc.enrollment.same.ruca))

x
PersonID
ps.in.same.ruca

So there are fewer observations in this dataset than previous subsets. Why? This is due to dropping a PersonID that wasn’t in the Southern graduate dataset. I did not do that for the previous subsets. However, those previous subsets will be filtered down once we join it with the master dataset.

This dataset provides the PersonID and whether they attended a post secondary institution in a location with the same, outside, or both outside and inside (if attended multiple post-secondary institutions) RUCA categories of their high school from which they graduated. There are 90,596 rows and 2 columns.

Up next we will create a dataset indicating whether a PersonID that graduated from a Southern MN high school attended a post-secondary institution in their high school’s EDR.

nsc.enrollment.same.edr <- nsc.enrollment.location.join %>%
  mutate(ps.in.same.edr = ifelse(ps.edr == grad.edr, "In same EDR", "Outside EDR"),
         ps.in.same.edr = ifelse(is.na(ps.in.same.edr), "Outside EDR", ps.in.same.edr)) %>%
  select(PersonID, ps.edr, grad.edr, ps.in.same.edr) %>%
  select(PersonID, ps.in.same.edr) %>%
  distinct(PersonID, ps.in.same.edr) %>%
  mutate(code = ifelse(ps.in.same.edr == "In same EDR", 1, 2)) %>%
  group_by(PersonID) %>%
  summarize(code = sum(code)) %>%
  ungroup() %>%
  mutate(ps.in.same.edr = ifelse(code == 1, "In same EDR",
                                  ifelse(code == 2, "Outside EDR", "Inside and outside same EDR"))) %>%
  select(PersonID, ps.in.same.edr)

kable(head(nsc.enrollment.same.edr))

PersonID	ps.in.same.edr
280	Outside EDR
301	Outside EDR
412	Outside EDR
513	In same EDR
687	Outside EDR
710	Outside EDR

kable(names(nsc.enrollment.same.edr))

x
PersonID
ps.in.same.edr

As expected there are 90,596 rows and 2 columns.

Next we will determine which Southern MN graduates attended a post-secondary institution in the same project region (Southern project region).

nsc.enrollment.same.pr <- nsc.enrollment.location.join %>%
  mutate(ps.in.same.pr = ifelse(ps.pr == grad.pr, "In same PR", "Outside PR"),
         ps.in.same.pr = ifelse(is.na(ps.in.same.pr), "Outside PR", ps.in.same.pr)) %>%
  select(PersonID, ps.pr, grad.pr, ps.in.same.pr) %>%
  select(PersonID, ps.in.same.pr) %>%
  distinct(PersonID, ps.in.same.pr) %>%
  mutate(code = ifelse(ps.in.same.pr == "In same PR", 1, 2)) %>%
  group_by(PersonID) %>%
  summarize(code = sum(code)) %>%
  ungroup() %>%
  mutate(ps.in.same.pr = ifelse(code == 1, "In same PR",
                                  ifelse(code == 2, "Outside PR", "Inside and outside same PR"))) %>%
  select(PersonID, ps.in.same.pr)

kable(head(nsc.enrollment.same.pr))

PersonID	ps.in.same.pr
280	Outside PR
301	Outside PR
412	In same PR
513	In same PR
687	Outside PR
710	Outside PR

kable(names(nsc.enrollment.same.pr))

x
PersonID
ps.in.same.pr

As expected, there are 90,596 rows and 2 columns. This dataset provides whether the post-secondary institution(s) attended were in the same planning region as the high school from which they graduates, outside of the planning region, or both (attended multiple institutions).

Next we want to see how many of the students leave the state to attend post-secondary education.

nsc.enrollment.in.MN <- nsc.enrollment.location.join %>%
  mutate(ps.in.MN = ifelse(ps.statefp == "27", "In MN", "Outside MN")) %>%
  select(PersonID, ps.in.MN) %>%
  distinct(PersonID, ps.in.MN) %>%
  mutate(code = ifelse(ps.in.MN == "In MN", 1, 2)) %>%
  group_by(PersonID) %>%
  summarize(code = sum(code)) %>%
  ungroup() %>%
  mutate(ps.in.MN = ifelse(code == 1, "In MN",
                                  ifelse(code == 2, "Outside MN", "Inside and outside MN"))) %>%
  select(PersonID, ps.in.MN)

kable(head(nsc.enrollment.in.MN))

PersonID	ps.in.MN
280	In MN
301	Outside MN
412	In MN
513	In MN
687	Outside MN
710	Outside MN

kable(names(nsc.enrollment.in.MN))

x
PersonID
ps.in.MN

As expected we have 90,596 rows and 2 columns. This dataset provides each distinct PersonID with whether they attended post secondary institutions inside MN, outside MN, or both.

Okay, now it’s time to join all of these with the master dataset. I will also create a new column that confirms whether they attended a college immediately (within the first year) after graduating high school.

master <- read_csv("Data/SLEDS/Masters/Master-8.csv") %>%
  mutate(PersonID = as.numeric(PersonID)) 

master.9 <- master %>%
  left_join(nsc.enrollment.confirm, by = "PersonID") %>%
  mutate(attended.ps = ifelse(is.na(attended.ps), "No", attended.ps)) %>%
  left_join(nsc.enrollment.first.attend, by = "PersonID") %>%
  mutate(attended.ps.first.year = as.integer(str_sub(first.attend.ps, 1, 4)),
         attended.ps.years.hsgrad = attended.ps.first.year - grad.year,
         attended.ps.within.first.year.hsgrad = ifelse(attended.ps.years.hsgrad < 2, "Yes", "No")) %>%
  left_join(nsc.enrollment.n.institutions, by = "PersonID") %>%
  left_join(nsc.enrollment.first.institution, by = "PersonID") %>%
  left_join(nsc.enrollment.institution.types, by = "PersonID") %>%
  left_join(nsc.enrollment.same.ruca, by = "PersonID") %>%
  left_join(nsc.enrollment.same.edr, by = "PersonID") %>%
  left_join(nsc.enrollment.same.pr, by = "PersonID") %>%
  left_join(nsc.enrollment.in.MN, by = "PersonID") %>%
  mutate(n.institutions = ifelse(is.na(n.institutions), 0, n.institutions),
         attended.ps.years.hsgrad = ifelse(attended.ps.years.hsgrad == -14, -1, attended.ps.years.hsgrad),
         Dem_Desc = fct_relevel(Dem_Desc, "Town/rural mix", "Urban/town/rural mix", "Entirely urban"))

nscenrollment.master <- read_csv("Data/SLEDS/Masters/Master-1.csv") %>%
  select(PersonID, grad.year) %>%
  left_join(nsc.enrollment.confirm, by = "PersonID") %>%
  mutate(attended.ps = ifelse(is.na(attended.ps), "No", attended.ps)) %>%
  left_join(nsc.enrollment.first.attend, by = "PersonID") %>%
  mutate(attended.ps.first.year = as.integer(str_sub(first.attend.ps, 1, 4)),
         attended.ps.years.hsgrad = attended.ps.first.year - grad.year,
         attended.ps.within.first.year.hsgrad = ifelse(attended.ps.years.hsgrad < 2, "Yes", "No")) %>%
  left_join(nsc.enrollment.n.institutions, by = "PersonID") %>%
  left_join(nsc.enrollment.first.institution, by = "PersonID") %>%
  left_join(nsc.enrollment.institution.types, by = "PersonID") %>%
  left_join(nsc.enrollment.same.ruca, by = "PersonID") %>%
  left_join(nsc.enrollment.same.edr, by = "PersonID") %>%
  left_join(nsc.enrollment.same.pr, by = "PersonID") %>%
  left_join(nsc.enrollment.in.MN, by = "PersonID") %>%
  mutate(n.institutions = ifelse(is.na(n.institutions), 0, n.institutions)) %>%
  select(-grad.year) 

write_csv(nscenrollment.master, "Data/SLEDS/Masters/nscenrollment_master.csv")

kable(head(master.9))

PersonID	K12OrganizationID	OrganizationName	county.name	countyfp	Dem_Desc	edr	grad.year	grad.year.covid	Gender	LimitedEnglishProficiencyIndicator	HomelessIndicator	economic.status	pseo.participant	SpecialEdStatus	non.english.home	RaceEthnicity	n.years.attended	ACTCompositeScore	ap.exam	total.cte.courses.taken	ACTE-SPED	Agriculture, Food, and Natural Resources	Architecture & Construction	Arts, A/V Technology & Communication	Business, Management & Administrative	Diversified	Education & Training	Finance	Government & Public Administration	Health Science	Hospitality & Tourism	Human Services	Information Technology	Law, Public Safety & Security	Manufacturing	Marketing	Program Area (Diversified)	Science, Technology, Engineering, & Mathematics	STEM	Transportation, Distribution & Logistics	Work Experience-Handicapped	Work Experience-Handicapped (16-20+ on IEP)	Work Experience/Career Exploration (age 14-15 on I	Work-Experience-Disadvantaged	Youth Apprenticeship	cte.achievement	english.learner	MCA.M	MCA.R	MCA.S	sat.taken	attended.ps	first.attend.ps	attended.ps.first.year	attended.ps.years.hsgrad	attended.ps.within.first.year.hsgrad	n.institutions	first.InstitutionSector	ps.sector.1	ps.sector.2	ps.sector.3	ps.sector.4	ps.sector.5	ps.sector.6	ps.sector.7	ps.sector.8	ps.sector.9	ps.in.same.ruca	ps.in.same.edr	ps.in.same.pr	ps.in.MN
7699638	29705	FILLMORE CENTRAL SENIOR HIGH	Fillmore	045	Urban/town/rural mix	EDR 10 - Southeast	2015	Pre-covid grad	M	N	N	0	1	0	0	White	3	19	0	2	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	CTE Participant	0	3	4	3	0	Yes	20150824	2015	0	Yes	3	1	1	0	0	1	0	0	0	0	0	Inside and outside same RUCA	Inside and outside same EDR	Inside and outside same PR	Inside and outside MN
2225025	132298	MAYO SENIOR HIGH	Olmsted	109	Entirely urban	EDR 10 - Southeast	2017	Pre-covid grad	M	N	N	0	0	0	0	White	3	29	0	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	No CTE	0	4	4	4	1	Yes	20170905	2017	0	Yes	1	1	1	0	0	0	0	0	0	0	0	In same RUCA	Outside EDR	Outside PR	In MN
10400093	62	Cannon Falls Secondary	Goodhue	049	Urban/town/rural mix	EDR 10 - Southeast	2014	Pre-covid grad	F	N	N	0	1	0	0	White	3	20	1	2	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	1	0	0	0	0	0	CTE Participant	0	4	4	3	0	Yes	20140825	2014	0	Yes	2	4	0	1	0	1	0	0	0	0	0	In same RUCA	Outside EDR	Outside PR	In MN
8786746	128661	LINCOLN SECONDARY	Wabasha	157	Urban/town/rural mix	EDR 10 - Southeast	2014	Pre-covid grad	F	N	N	0	1	0	0	White	3	24	1	2	0	0	0	0	0	1	0	0	0	0	0	0	0	0	1	0	0	0	0	0	0	0	0	0	0	CTE Participant	0	4	4	3	0	Yes	20140902	2014	0	Yes	1	2	0	1	0	0	0	0	0	0	0	In same RUCA	Outside EDR	In same PR	In MN
11122574	109999	AUSTIN SENIOR HIGH	Mower	099	Urban/town/rural mix	EDR 10 - Southeast	2011	Pre-covid grad	F	N	N	0	1	0	0	White	3	19	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	No CTE	0	3	4	3	0	Yes	20110822	2011	0	Yes	2	4	1	0	0	1	0	0	0	0	0	In same RUCA	Inside and outside same EDR	Inside and outside same PR	In MN
10181573	155484	CHATFIELD SECONDARY	Olmsted	109	Entirely urban	EDR 10 - Southeast	2010	Pre-covid grad	M	N	N	0	NA	0	0	White	3	33	1	1	0	0	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	CTE Participant	0	4	4	4	0	Yes	20120827	2012	2	No	2	4	1	0	0	1	0	0	0	0	0	In same RUCA	Outside EDR	Outside PR	In MN

kable(names(master.9))

x
PersonID
K12OrganizationID
OrganizationName
county.name
countyfp
Dem_Desc
edr
grad.year
grad.year.covid
Gender
LimitedEnglishProficiencyIndicator
HomelessIndicator
economic.status
pseo.participant
SpecialEdStatus
non.english.home
RaceEthnicity
n.years.attended
ACTCompositeScore
ap.exam
total.cte.courses.taken
ACTE-SPED
Agriculture, Food, and Natural Resources
Architecture & Construction
Arts, A/V Technology & Communication
Business, Management & Administrative
Diversified
Education & Training
Finance
Government & Public Administration
Health Science
Hospitality & Tourism
Human Services
Information Technology
Law, Public Safety & Security
Manufacturing
Marketing
Program Area (Diversified)
Science, Technology, Engineering, & Mathematics
STEM
Transportation, Distribution & Logistics
Work Experience-Handicapped
Work Experience-Handicapped (16-20+ on IEP)
Work Experience/Career Exploration (age 14-15 on I
Work-Experience-Disadvantaged
Youth Apprenticeship
cte.achievement
english.learner
MCA.M
MCA.R
MCA.S
sat.taken
attended.ps
first.attend.ps
attended.ps.first.year
attended.ps.years.hsgrad
attended.ps.within.first.year.hsgrad
n.institutions
first.InstitutionSector
ps.sector.1
ps.sector.2
ps.sector.3
ps.sector.4
ps.sector.5
ps.sector.6
ps.sector.7
ps.sector.8
ps.sector.9
ps.in.same.ruca
ps.in.same.edr
ps.in.same.pr
ps.in.MN

As expected, we have the same number of rows as the original master dataset - 122,938 rows. Here are the explanations of all the new columns added to the master dataset.

• attended.ps - this is a confirmation that the PersonID attended a post-secondary institution at some point after graduating high school.
• attended.ps.years.hsgrad - the number of years after graduation that a PersonID attended a post-secondary education. In some cases, the value is negative indicating that the PersonID attended a post-secondary institution before graduating high school.
• attended.ps.within.first.year.hsgrad - did the PersonID attend a post-secondary education institution within 1 year or less from graduating high school.
• n.institutions - how many post-secondary institutions the PersonID attended
• first.InstitutionSector - the sector of the first post-secondary institution attended.
• InstitutionSector - The sector of the post-secondary institution attended (could be multiple)
• ps.in.same.ruca - did the PersonID attend a post-secondary institution in the same RUCA category as their high school
• ps.in.same.edr - did the PersonID attend a post-secondary institution in the same economic development region as high school
• ps.in.same.pr - did the PersonID attend a post-secondary institution in the same planning region as high school
• ps.in.MN = did the PersonID attend a post-secondary institution in Minnesota.

Summary of students attending post-secondary

Lets summarize the percentage of students that attended a post-secondary institution and compare across RUCA categories.

Below is the percentage of students that attended a post-secondary institution from the entire dataset. 72% of the PersonIDs in the dataset attended a post-secondary institution at some point between 2007 and 2024.

attending.ps.pct.total <- master.9 %>%
  group_by(attended.ps) %>%
  summarize(n = n()) %>%
  ungroup() %>%
  mutate(pct = n / sum(n)) 

datatable(attending.ps.pct.total, class = "cell-border stripe", filter = "top", rownames = FALSE,
          options = list(columnDefs = list(list(className = "dt-center", targets = 1:2)))) %>%
  formatCurrency(2, "", digits = 0) %>%
  formatPercentage(3)

Now lets check to see if this percentage is statistically significantly different by RUCA group.

The crosstabs do indicate that there is a statistically significant difference in the percentage of graduates that attend a post-secondary institution by high school RUCA category. Entirely urban districts have the highest percentage of graduates attend a post-secondary institution - 75% compared to 70% for the other RUCA categories.

Chart - RUCA

attending.ps.pct.ruca <- master.9 %>%
  group_by(Dem_Desc, attended.ps) %>%
  summarize(n = n()) %>%
  ungroup() %>%
  group_by(Dem_Desc) %>%
  mutate(pct = n / sum(n)) %>%
  ungroup() %>%
  mutate(data_id = as.character(seq(n())))

attending.ps.pct.ruca.plot <- ggplot(attending.ps.pct.ruca, aes(Dem_Desc, pct, fill = attended.ps, group = attended.ps)) +
    geom_col_interactive(position = "dodge", aes(data_id = data_id, tooltip = paste(Dem_Desc, "\nAttended a post-secondary intitution: ", attended.ps, "\nNumber: ", comma(n, accuracy = 1), "\nPercent of students: ", percent(pct, accuracy = .1), sep = ""))) +
    geom_label(aes(label = percent(pct, accuracy = .1),), position = position_dodge(width = .9), show.legend = FALSE, color = "white", size = 5) +
    labs(x="", y = "", color="", title = "Attended a post-secondary institution")+
    scale_y_continuous(labels=scales::percent)+
    scale_x_discrete(guide = guide_axis(n.dodge = 1)) +
    theme_bar+
    scale_fill_manual(values = brewer.pal(n = 5, "RdYlBu"),
                       guide = guide_legend(ncol = 3)) +
    theme(legend.position = "bottom",
          text = element_text(size = 16))


girafe(ggobj = attending.ps.pct.ruca.plot) %>%
  girafe_options(opts_selection(type = "none"),
                 opts_sizing(rescale = FALSE))

Cross-tabs - RUCA

CrossTable(master.9$Dem_Desc, master.9$attended.ps, expected = TRUE, prop.t = FALSE, prop.c = FALSE, prop.chisq = FALSE)

 
   Cell Contents
|-------------------------|
|                       N |
|              Expected N |
|           N / Row Total |
|-------------------------|

 
Total Observations in Table:  122938 

 
                     | master.9$attended.ps 
   master.9$Dem_Desc |        No |       Yes | Row Total | 
---------------------|-----------|-----------|-----------|
      Town/rural mix |      6322 |     15801 |     22123 | 
                     |  6129.722 | 15993.278 |           | 
                     |     0.286 |     0.714 |     0.180 | 
---------------------|-----------|-----------|-----------|
Urban/town/rural mix |     17175 |     41824 |     58999 | 
                     | 16347.126 | 42651.874 |           | 
                     |     0.291 |     0.709 |     0.480 | 
---------------------|-----------|-----------|-----------|
      Entirely urban |     10566 |     31250 |     41816 | 
                     | 11586.152 | 30229.848 |           | 
                     |     0.253 |     0.747 |     0.340 | 
---------------------|-----------|-----------|-----------|
        Column Total |     34063 |     88875 |    122938 | 
---------------------|-----------|-----------|-----------|

 
Statistics for All Table Factors


Pearson's Chi-squared test 
------------------------------------------------------------
Chi^2 =  190.5888     d.f. =  2     p =  4.113063e-42 


 

Summary of years between graduation and attending post-secondary

Now lets take a look at the number of years between high school graduation and attending post-secondary. We will begin with the total number of students before diving into differences across RUCA categories and regions.

A huge majority of students that attended a post secondary institution waited less than 1 year after graduating high school (87%)

Summary Stats

years.between.grad.ps.total <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  select(PersonID, first.attend.ps, attended.ps.first.year, attended.ps.within.first.year.hsgrad, attended.ps.years.hsgrad, grad.year) %>%
  summarize(`Number attended post-secondary` = comma(n()),
            `Mean years` = mean(attended.ps.years.hsgrad),
            `Median years` = median(attended.ps.years.hsgrad),
            `Max years` = max(attended.ps.years.hsgrad),
            `Min years` = min(attended.ps.years.hsgrad),
            `Standard deviation` = sd(attended.ps.years.hsgrad)) 

datatable(years.between.grad.ps.total, class = "cell-border stripe", filter = "top", rownames = FALSE,
          options = list(columnDefs = list(list(className = "dt-center", targets = 0:5)))) %>%
  formatCurrency(2, "", digits = 3) %>%
  formatCurrency(6, "", digits = 4)

Distribution

years.between.grad.ps.dist.total <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  mutate(attended.ps.years.hsgrad.bins = cut(attended.ps.years.hsgrad,
                                             breaks = c(-1.1, .1, 1, 2, 3, 100),
                                             labels = c("Less than one year", "1 year", "2 years", "3 years", "4 years or more")),
         data_id = as.character(seq(n()))) %>%
  group_by(attended.ps.years.hsgrad.bins) %>%
  summarize(n = n()) %>%
  ungroup() %>%
  mutate(pct = n / sum(n),
         data_id = as.character(seq(n())))
  

years.between.grad.ps.dist.total.plot <- ggplot(years.between.grad.ps.dist.total, aes(attended.ps.years.hsgrad.bins, pct)) +
    geom_col_interactive(aes(data_id = data_id, tooltip = paste("\nNumber of years between high school graduation and attending post secondary: ", attended.ps.years.hsgrad.bins, "\nNumber of students: ", comma(n, accuracy = 1), "\nPercent of students: ", percent(pct, accuracy = .1), sep = ""))) +
    geom_label(aes(label = percent(pct, accuracy = .1)), show.legend = FALSE, color =  "black", size = 5) +
    labs(x="", y = "", color="", title = "Percent of students and the number of years between high school and attending post secondary")+
    scale_y_continuous(labels=scales::percent)+
    scale_x_discrete(guide = guide_axis(n.dodge = 1)) +
    theme_bar+
  theme(legend.position = "none",
          text = element_text(size = 16),
        axis.text.x = element_text(angle = 25, hjust = .9))

girafe(ggobj = years.between.grad.ps.dist.total.plot) %>%
  girafe_options(opts_selection(type = "none"),
                 opts_sizing(rescale = FALSE))

Summary of number of colleges attended

Next we will summarize the number of colleges attended by producing the summary statistics and distribution.

The table and distribution chart below show that a large majority (60%) of individuals attended only one post secondary institutions followed by 27.4% attending two.

Summary Stats

n.colleges.total <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  summarize(`Number` = n(),
            `Mean` = mean(n.institutions),
            `Median` = median(n.institutions),
            `Max` = max(n.institutions),
            `Min` = min(n.institutions),
            `Standard deviation` = sd(n.institutions))

datatable(n.colleges.total, class = "cell-border stripe", filter = "top", rownames = FALSE,
          options = list(columnDefs = list(list(className = "dt-center", targets = 1:5)))) %>%
  formatCurrency(2, "", digits = 2) %>%
  formatCurrency(6, "", digits = 2)

Distribution

n.colleges.dist.total <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  mutate(n.institutions.bins = cut(n.institutions,
                               breaks = c(0, 1, 2, 3, 4, 20),
                               labels = c("One", "Two", "Three", "Four", "More than 4"))) %>%
  group_by(n.institutions) %>%
  summarise(n = n()) %>%
  ungroup() %>%
  mutate(pct = n / sum(n),
         data_id = as.character(seq(n())))


n.colleges.dist.total.plot <- ggplot(n.colleges.dist.total, aes(n.institutions, pct)) +
  geom_col_interactive(aes(data_id = data_id, tooltip = paste("Number of institutions attended: ", comma(n.institutions, accuracy = 1), "\nNumber of students: ", comma(n, accuracy = 1), "\nPercent of students: ", percent(pct, accuracy = .1), sep = ""))) +
  geom_label(aes(label = percent(pct, accuracy = .1)), show.legend = FALSE, color = "black", size = 5) +
  labs(x="", y = "", color="", title = "Percent of students by number of post secondary institutions attended")+
    scale_y_continuous(labels=scales::percent)+
    theme_bar+
    theme(legend.position = "none",
          text = element_text(size = 16))


girafe(ggobj = n.colleges.dist.total.plot) %>%
  girafe_options(opts_selection(type = "none"),
                 opts_sizing(rescale = FALSE))

I’m not sure it’s all that important knowing whether there are differences in the percentage of students and the number of institutions they’ve attended across RUCA categories or regions. So we will skip that analysis for now.

Summary of types of first college attended

I think this will be very interesting - we want to see what they breakdown is of students attending different types of colleges. We will start by summarizing the total dataset before looking at differences across RUCA categories and regions.

The chart below provides the percentage of students in the entire dataset that attended each institution sector. By far, a huge majority attend either a public 4-year (44%) or public 2-year (38%).

sectors <- read_xlsx('/Users/kellyasche/Library/CloudStorage/GoogleDrive-kasche@ruralmn.org/My Drive/Research/FY25-27/FY26/OHE Journey to meaningful employment/Data/SLEDS/NSC/IPEDS/institution-sectors.xlsx')

college.type.summary.total <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  group_by(first.InstitutionSector) %>%
  summarize(n = n()) %>%
  ungroup() %>%
  mutate(pct = n / sum(n)) %>%
  left_join(sectors, by = c("first.InstitutionSector" = "Code")) %>%
  mutate(data_id = as.character(seq(n())))

college.type.summary.total.plot <- ggplot(college.type.summary.total, aes(x = fct_rev(fct_reorder(Sector, pct)), pct)) +
    geom_col_interactive(aes(data_id = data_id, tooltip = paste("College sector: ", Sector, "\nNumber of students in college sector: ", comma(n, accuracy = 1), "\nPercent of students in college sector: ", percent(pct, accuracy = .1), sep = ""))) +
    geom_label(aes(label = percent(pct, accuracy = .1)), show.legend = FALSE, color = "black", size = 5) +
    labs(x="", y = "", color="", title = "Percent of students first college sector")+
    scale_y_continuous(labels=scales::percent)+
    scale_x_discrete(guide = guide_axis(n.dodge = 1)) +
    theme_bar+
    theme(legend.position = "none",
          text = element_text(size = 18),
          axis.text.x = element_text(angle = 45, hjust = 1))


girafe(ggobj = college.type.summary.total.plot, width_svg = 10, height_svg = 6) %>%
  girafe_options(opts_selection(type = "none"),
                 opts_sizing(rescale = FALSE))

Next we will check to see if those percentages are significantly different by RUCA category of the graduates high school.

The crosstabs indicate that there is a significant difference in the percentage of students attending different institution sectors depending on the RUCA category of their high school. The p-value was near zero.

The primary difference is the graduates from town/rural mix districts attend 2-year colleges at a higher percentage and attend public, 4-year colleges at a lower level.

Chart - RUCA

first.college.sector.ruca <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  group_by(Dem_Desc, first.InstitutionSector) %>%
  summarize(n = n()) %>%
  ungroup() %>%
  group_by(Dem_Desc) %>%
  mutate(pct = n / sum(n)) %>%
  ungroup() %>%
  left_join(sectors, by = c("first.InstitutionSector" = "Code")) %>%
  mutate(data_id = as.character(seq(n()))) %>%
  mutate(total.pct = n / sum(n))

first.college.sector.ruca.plot <- ggplot(first.college.sector.ruca, aes(fct_rev(fct_reorder(Sector, pct)), pct, fill = Dem_Desc, group = Dem_Desc)) +
    geom_col_interactive(position = "dodge", aes(data_id = data_id, tooltip = paste(Dem_Desc, "\nInstitution sector: ", Sector, "\nNumber of graduates attending: ", comma(n, accuracy = 1), "\nPercent of students attending: ", percent(pct, accuracy = .1), sep = ""))) +
    geom_label(position = position_dodge(width = .9), aes(label = percent(pct, accuracy = .1)), show.legend = FALSE, color = "black", size = 5) +
    labs(x="", y = "", color="", title = "Percent of students and first college attended\nby RUCA category")+
    scale_y_continuous(labels=scales::percent)+
    theme_bar+
    scale_fill_manual(values= color.ruca,
                       guide = guide_legend(ncol = 3)) +
    theme(legend.position = "bottom",
          text = element_text(size = 16),
          axis.text.x = element_text(angle = 45, hjust = 1))


girafe(ggobj = first.college.sector.ruca.plot) %>%
  girafe_options(opts_selection(type = "none"),
                 opts_sizing(rescale = FALSE))

Cross-tabs - RUCA

first.college.sector.ct.1.ruca <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  select(Dem_Desc, first.InstitutionSector) %>%
  left_join(sectors, by = c("first.InstitutionSector" = "Code")) %>%
  filter(Sector %in% c("Public, 4-year or above", "Public, 2-year", "Private not-for-profit, 4-year or above", "Private for-profit, 4-year or above")) %>%
  mutate(Sector = as_factor(Sector))

CrossTable(first.college.sector.ct.1.ruca$Dem_Desc, first.college.sector.ct.1.ruca$first.InstitutionSector, expected = TRUE, prop.t = FALSE, prop.c = FALSE, prop.chisq = FALSE)

 
   Cell Contents
|-------------------------|
|                       N |
|              Expected N |
|           N / Row Total |
|-------------------------|

 
Total Observations in Table:  88752 

 
                                        | first.college.sector.ct.1.ruca$first.InstitutionSector 
first.college.sector.ct.1.ruca$Dem_Desc |         1 |         2 |         3 |         4 | Row Total | 
----------------------------------------|-----------|-----------|-----------|-----------|-----------|
                         Town/rural mix |      6899 |      2456 |       262 |      6162 |     15779 | 
                                        |  6880.022 |  2632.678 |   195.211 |  6071.089 |           | 
                                        |     0.437 |     0.156 |     0.017 |     0.391 |     0.178 | 
----------------------------------------|-----------|-----------|-----------|-----------|-----------|
                   Urban/town/rural mix |     17411 |      7320 |       503 |     16514 |     41748 | 
                                        | 18203.129 |  6965.526 |   516.488 | 16062.857 |           | 
                                        |     0.417 |     0.175 |     0.012 |     0.396 |     0.470 | 
----------------------------------------|-----------|-----------|-----------|-----------|-----------|
                         Entirely urban |     14388 |      5032 |       333 |     11472 |     31225 | 
                                        | 13614.849 |  5209.796 |   386.302 | 12014.054 |           | 
                                        |     0.461 |     0.161 |     0.011 |     0.367 |     0.352 | 
----------------------------------------|-----------|-----------|-----------|-----------|-----------|
                           Column Total |     38698 |     14808 |      1098 |     34148 |     88752 | 
----------------------------------------|-----------|-----------|-----------|-----------|-----------|

 
Statistics for All Table Factors


Pearson's Chi-squared test 
------------------------------------------------------------
Chi^2 =  183.4382     d.f. =  6     p =  6.313396e-37 


 

Next we will check to see if those percentages are significantly different by EDR of the graduates high school.

The crosstabs indicate that there is a significant difference in the percentage of students attending different institution sectors depending on the EDR of their high school. The p-value was near zero.

The biggest difference is that EDR 9 attends 4-year public institution at a much higher rate while attending a 2-year public at a much lower rate. EDR 10 was the opposite.

Chart - EDR

first.college.sector.edr <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  group_by(edr, first.InstitutionSector) %>%
  summarize(n = n()) %>%
  ungroup() %>%
  group_by(edr) %>%
  mutate(pct = n / sum(n)) %>%
  ungroup() %>%
  left_join(sectors, by = c("first.InstitutionSector" = "Code")) %>%
  mutate(data_id = as.character(seq(n()))) %>%
  mutate(total.pct = n / sum(n))

first.college.sector.edr.plot <- ggplot(first.college.sector.edr, aes(fct_rev(fct_reorder(Sector, pct)), pct, fill = edr, group = edr)) +
    geom_col_interactive(position = "dodge", aes(data_id = data_id, tooltip = paste(edr, "\nInstitution sector: ", Sector, "\nNumber of graduates attending: ", comma(n, accuracy = 1), "\nPercent of students attending: ", percent(pct, accuracy = .1), sep = ""))) +
    geom_label(position = position_dodge(width = .9), aes(label = percent(pct, accuracy = .1)), show.legend = FALSE, color = "black", size = 5) +
    labs(x="", y = "", color="", title = "Percent of students and first college attended\nby edr category")+
    scale_y_continuous(labels=scales::percent)+
    theme_bar+
    scale_fill_manual(values= color.edr,
                       guide = guide_legend(ncol = 3)) +
    theme(legend.position = "bottom",
          text = element_text(size = 16),
          axis.text.x = element_text(angle = 45, hjust = 1))


girafe(ggobj = first.college.sector.edr.plot, height_svg = 10) %>%
  girafe_options(opts_selection(type = "none"),
                 opts_sizing(rescale = FALSE))

Cross-tabs - edr

first.college.sector.ct.1.edr <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  select(edr, first.InstitutionSector) %>%
  left_join(sectors, by = c("first.InstitutionSector" = "Code")) %>%
  filter(Sector %in% c("Public, 4-year or above", "Public, 2-year", "Private not-for-profit, 4-year or above", "Private for-profit, 4-year or above")) %>%
  mutate(Sector = as_factor(Sector))

CrossTable(first.college.sector.ct.1.edr$edr, first.college.sector.ct.1.edr$first.InstitutionSector, expected = TRUE, prop.t = FALSE, prop.c = FALSE, prop.chisq = FALSE)

 
   Cell Contents
|-------------------------|
|                       N |
|              Expected N |
|           N / Row Total |
|-------------------------|

 
Total Observations in Table:  88752 

 
                                  | first.college.sector.ct.1.edr$first.InstitutionSector 
first.college.sector.ct.1.edr$edr |         1 |         2 |         3 |         4 | Row Total | 
----------------------------------|-----------|-----------|-----------|-----------|-----------|
               EDR 10 - Southeast |     26359 |     10619 |       686 |     25620 |     63284 | 
                                  | 27593.341 | 10558.742 |   782.921 | 24348.995 |           | 
                                  |     0.417 |     0.168 |     0.011 |     0.405 |     0.713 | 
----------------------------------|-----------|-----------|-----------|-----------|-----------|
            EDR 9 - South Central |     12339 |      4189 |       412 |      8528 |     25468 | 
                                  | 11104.659 |  4249.258 |   315.079 |  9799.005 |           | 
                                  |     0.484 |     0.164 |     0.016 |     0.335 |     0.287 | 
----------------------------------|-----------|-----------|-----------|-----------|-----------|
                     Column Total |     38698 |     14808 |      1098 |     34148 |     88752 | 
----------------------------------|-----------|-----------|-----------|-----------|-----------|

 
Statistics for All Table Factors


Pearson's Chi-squared test 
------------------------------------------------------------
Chi^2 =  466.635     d.f. =  3     p =  8.106745e-101 


 

Summary of attending college with same RUCA category

Up next is determining how many students attended a post-secondary institution located in a county with the same RUCA category as their high school. First we will look at the percentages of the total dataset and then we will break it up.

The chart below shows that 37% of students graduating from a Southern MN high school attended a post secondary institution that was located in a county with the same RUCA category as their high school.

ps.same.ruca.total <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  group_by(ps.in.same.ruca) %>%
  summarize(n = n()) %>%
  ungroup() %>%
  mutate(pct = n / sum(n),
         data_id = as.character(seq(n())))

ps.same.ruca.total.plot <- ggplot(ps.same.ruca.total, aes(fct_rev(fct_reorder(ps.in.same.ruca, pct)), pct)) +
    geom_col_interactive(aes(data_id = data_id, tooltip = paste("Location of post-secondary institution: ", ps.in.same.ruca, "\nNumber of students: ", comma(n, accuracy = 1), "\nPercent of students: ", percent(pct, accuracy = .1), sep = ""))) +
    geom_label(aes(label = percent(pct, accuracy = .1)), show.legend = FALSE, color = "black", size = 5) +
    labs(x="", y = "", color="", title = "Percent of students attending post secondary in RUCA category")+
    scale_y_continuous(labels=scales::percent)+
    scale_x_discrete(guide = guide_axis(n.dodge = 1)) +
    theme_bar+
    theme(legend.position = "none",
          text = element_text(size = 18))


girafe(ggobj = ps.same.ruca.total.plot, width_svg = 10, height_svg = 6) %>%
  girafe_options(opts_selection(type = "none"),
                 opts_sizing(rescale = FALSE))

Next, lets check to see if the type of RUCA category they graduated from is related to whether the college they attend is in the same or different RUCA category.

The crosstabs indicate that there is a relationship in the location of the individuals high school graduation and the RUCA category of their post-secondary institution(s). The p-value was 0.

A significantly higher percentage of town/rural mix graduates attended a post-secondary institution outside their RUCA category.

Chart - RUCA

ps.same.ruca.ruca <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  group_by(Dem_Desc, ps.in.same.ruca) %>%
  summarise(n = n()) %>%
  ungroup() %>%
  complete(ps.in.same.ruca, Dem_Desc) %>%
  mutate(n = ifelse(is.na(n), 0, n)) %>%
  group_by(Dem_Desc) %>%
  mutate(pct = n / sum(n)) %>%
  ungroup() %>%
  mutate(data_id = as.character(seq(n())))

ps.same.ruca.ruca.plot <- ggplot(ps.same.ruca.ruca, aes(ps.in.same.ruca, pct, group = Dem_Desc, fill = Dem_Desc)) +
    geom_col_interactive(position = "dodge", aes(data_id = data_id, tooltip = paste("RUCA category comparison between high school and college: ", "\nNumber of students: ", comma(n, accuracy = 1), "\nPercent of students: ", percent(pct, accuracy = .1), sep = ""))) +
    geom_label(position = position_dodge(width = .9), aes(label = percent(pct, accuracy = .1)), show.legend = FALSE, color = "black", size = 5) +
    labs(x="", y = "", color="", title = "Percent of students attending post-secondary of same RUCA\ncategory by RUCA category of high school")+
    scale_y_continuous(labels=scales::percent)+
    theme_bar+
    scale_fill_manual(values= color.ruca,
                       guide = guide_legend(ncol = 3),
                      breaks = c("Entirely urban", "Town/rural mix", "Urban/town/rural mix")) +
    theme(legend.position = "bottom",
          text = element_text(size = 16),
          axis.text.x = element_text(angle = 25, hjust = .9))


girafe(ggobj = ps.same.ruca.ruca.plot) %>%
  girafe_options(opts_selection(type = "none"),
                 opts_sizing(rescale = FALSE))

Cross-tabs - RUCA

ps.same.ruca.ct.ruca <- master.9 %>%
  filter(attended.ps == "Yes")

CrossTable(ps.same.ruca.ct.ruca$Dem_Desc, ps.same.ruca.ct.ruca$ps.in.same.ruca, expected = TRUE, prop.t = FALSE, prop.c = FALSE, prop.chisq = FALSE)

 
   Cell Contents
|-------------------------|
|                       N |
|              Expected N |
|           N / Row Total |
|-------------------------|

 
Total Observations in Table:  88868 

 
                              | ps.same.ruca.ct.ruca$ps.in.same.ruca 
ps.same.ruca.ct.ruca$Dem_Desc |                 In same RUCA | Inside and outside same RUCA |                 Outside RUCA |                    Row Total | 
------------------------------|------------------------------|------------------------------|------------------------------|------------------------------|
               Town/rural mix |                         2338 |                         1986 |                        11477 |                        15801 | 
                              |                     5793.712 |                     3565.840 |                     6441.448 |                              | 
                              |                        0.148 |                        0.126 |                        0.726 |                        0.178 | 
------------------------------|------------------------------|------------------------------|------------------------------|------------------------------|
         Urban/town/rural mix |                        18267 |                         9940 |                        13614 |                        41821 | 
                              |                    15334.398 |                     9437.820 |                    17048.782 |                              | 
                              |                        0.437 |                        0.238 |                        0.326 |                        0.471 | 
------------------------------|------------------------------|------------------------------|------------------------------|------------------------------|
               Entirely urban |                        11980 |                         8129 |                        11137 |                        31246 | 
                              |                    11456.890 |                     7051.341 |                    12737.769 |                              | 
                              |                        0.383 |                        0.260 |                        0.356 |                        0.352 | 
------------------------------|------------------------------|------------------------------|------------------------------|------------------------------|
                 Column Total |                        32585 |                        20055 |                        36228 |                        88868 | 
------------------------------|------------------------------|------------------------------|------------------------------|------------------------------|

 
Statistics for All Table Factors


Pearson's Chi-squared test 
------------------------------------------------------------
Chi^2 =  8366.953     d.f. =  4     p =  0 


 

Next, lets check to see if the EDR they graduated from is related to whether the college they attend is in the same or different EDR.

The crosstabs indicate that there is a relationship in the location of the individuals high school graduation and the EDR their post-secondary institution(s). The p-value was 0.

The differences are minimal.

Chart - EDR

ps.same.edr.edr <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  group_by(edr, ps.in.same.edr) %>%
  summarise(n = n()) %>%
  ungroup() %>%
  complete(ps.in.same.edr, edr) %>%
  mutate(n = ifelse(is.na(n), 0, n)) %>%
  group_by(edr) %>%
  mutate(pct = n / sum(n)) %>%
  ungroup() %>%
  mutate(data_id = as.character(seq(n())))

ps.same.edr.edr.plot <- ggplot(ps.same.edr.edr, aes(ps.in.same.edr, pct, group = edr, fill = edr)) +
    geom_col_interactive(position = "dodge", aes(data_id = data_id, tooltip = paste("EDR comparison between high school and college: ", "\nNumber of students: ", comma(n, accuracy = 1), "\nPercent of students: ", percent(pct, accuracy = .1), sep = ""))) +
    geom_label(position = position_dodge(width = .9), aes(label = percent(pct, accuracy = .1)), show.legend = FALSE, color = "black", size = 5) +
    labs(x="", y = "", color="", title = "Percent of students attending post-secondary of same EDR\nby EDR of high school")+
    scale_y_continuous(labels=scales::percent)+
    theme_bar+
    scale_fill_manual(values= color.edr,
                       guide = guide_legend(ncol = 3)) +
    theme(legend.position = "bottom",
          text = element_text(size = 16),
          axis.text.x = element_text(angle = 25, hjust = .9))


girafe(ggobj = ps.same.edr.edr.plot) %>%
  girafe_options(opts_selection(type = "none"),
                 opts_sizing(rescale = FALSE))

Cross-tabs - EDR

ps.same.ruca.ct.edr <- master.9 %>%
  filter(attended.ps == "Yes")

CrossTable(ps.same.ruca.ct.edr$edr, ps.same.ruca.ct.edr$ps.in.same.edr, expected = TRUE, prop.t = FALSE, prop.c = FALSE, prop.chisq = FALSE)

 
   Cell Contents
|-------------------------|
|                       N |
|              Expected N |
|           N / Row Total |
|-------------------------|

 
Total Observations in Table:  88875 

 
                        | ps.same.ruca.ct.edr$ps.in.same.edr 
ps.same.ruca.ct.edr$edr |                 In same EDR | Inside and outside same EDR |                 Outside EDR |                   Row Total | 
------------------------|-----------------------------|-----------------------------|-----------------------------|-----------------------------|
     EDR 10 - Southeast |                       17127 |                       12207 |                       34036 |                       63370 | 
                        |                   17623.812 |                   12196.274 |                   33549.914 |                             | 
                        |                       0.270 |                       0.193 |                       0.537 |                       0.713 | 
------------------------|-----------------------------|-----------------------------|-----------------------------|-----------------------------|
  EDR 9 - South Central |                        7590 |                        4898 |                       13017 |                       25505 | 
                        |                    7093.188 |                    4908.726 |                   13503.086 |                             | 
                        |                       0.298 |                       0.192 |                       0.510 |                       0.287 | 
------------------------|-----------------------------|-----------------------------|-----------------------------|-----------------------------|
           Column Total |                       24717 |                       17105 |                       47053 |                       88875 | 
------------------------|-----------------------------|-----------------------------|-----------------------------|-----------------------------|

 
Statistics for All Table Factors


Pearson's Chi-squared test 
------------------------------------------------------------
Chi^2 =  73.37579     d.f. =  2     p =  1.165866e-16 


 

Summary of attending college in same planning region as high school

Now we want to see how many students attend a college that is located in the same planning region as their high school.

The chart below shows that nearly 45% of students graduating from Southern MN High schools leave the region to attend post secondary education.

ps.same.pr.total <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  group_by(ps.in.same.pr) %>%
  summarize(n = n()) %>%
  ungroup() %>%
  mutate(pct = n / sum(n),
         data_id = as.character(seq(n())))

ps.same.pr.total.plot <- ggplot(ps.same.pr.total, aes(fct_rev(fct_reorder(ps.in.same.pr, pct)), pct)) +
    geom_col_interactive(aes(data_id = data_id, tooltip = paste("College planning region location: ", ps.in.same.pr, "\nNumber of students: ", comma(n, accuracy = 1), "\nPercent of students: ", percent(pct, accuracy = .1), sep = ""))) +
    geom_label(aes(label = percent(pct, accuracy = .1)), show.legend = FALSE, color = "black", size = 5) +
    labs(x="", y = "", color="", title = "Percent of students attending college within\nsame planning region")+
    scale_y_continuous(labels=scales::percent)+
    theme_bar+
    theme(legend.position = "none",
          text = element_text(size = 16),
          axis.text.x = element_text(angle = 25, hjust = .9))


girafe(ggobj = ps.same.pr.total.plot) %>%
  girafe_options(opts_selection(type = "none"),
                 opts_sizing(rescale = FALSE))

Summary of attending college in Minnesota

Now we want to see how many students stay or leave Minnesota to attend college.

The chart below shows that 62% of students graduating from Southern MN high schools attend a post secondary institution in Minnesota.

ps.in.MN.total <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  group_by(ps.in.MN) %>%
  summarize(n = n()) %>%
  ungroup() %>%
  mutate(pct = n / sum(n),
         data_id = as.character(seq(n())))

ps.in.MN.total.plot <- ggplot(ps.in.MN.total, aes(fct_rev(fct_reorder(ps.in.MN, pct)), pct)) +
    geom_col_interactive(aes(data_id = data_id, tooltip = paste("College planning region location: ", ps.in.MN, "\nNumber of students: ", comma(n, accuracy = 1), "\nPercent of students: ", percent(pct, accuracy = .1), sep = ""))) +
    geom_label(aes(label = percent(pct, accuracy = .1)), show.legend = FALSE, color = "black", size = 5) +
    labs(x="", y = "", color="", title = "Percent of students attending college in\nMinnesota")+
    scale_y_continuous(labels=scales::percent)+
    scale_x_discrete(guide = guide_axis(n.dodge = 1)) +
    theme_bar+
    theme(legend.position = "none",
          text = element_text(size = 16),
          axis.text.x = element_text(angle = 25, hjust = .9))


girafe(ggobj = ps.in.MN.total.plot) %>%
  girafe_options(opts_selection(type = "none"),
                 opts_sizing(rescale = FALSE))

Next, lets check to see if the RUCA category of their high school is related to whether they attend a college inside or outside Minnesota.

The crosstabs below indicate that there is a relationship between the RUCA category of a student’s high school and whether they attend a college inside or outside of Minnesota. The p-value was near zero.

The differences are minimal.

Chart - RUCA

ps.in.MN.ruca <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  group_by(Dem_Desc, ps.in.MN) %>%
  summarize(n = n()) %>%
  ungroup() %>%
  group_by(Dem_Desc) %>%
  mutate(pct = n / sum(n)) %>%
  ungroup() %>%
  mutate(data_id = as.character(seq(n())))

ps.in.MN.ruca.plot <- ggplot(ps.in.MN.ruca, aes(ps.in.MN, pct, fill = Dem_Desc, group = Dem_Desc)) +
    geom_col_interactive(position = "dodge", aes(data_id = data_id, tooltip = paste(Dem_Desc, "\nCollege in Minnesota or not: ", ps.in.MN, "\nNumber of students: ", comma(n, accuracy = 1), "\nPercent of students: ", percent(pct, accuracy = .1), sep = ""))) +
    geom_label(position = position_dodge(width = .9), aes(label = percent(pct, accuracy = .1)), show.legend = FALSE, color = "black", size = 5) +
    labs(x="", y = "", color="", title = "Percent of students attending college in\nMinnesota")+
    scale_y_continuous(labels=scales::percent)+
    theme_bar+
    scale_fill_manual(values= color.ruca,
                       guide = guide_legend(ncol = 3)) +
    theme(legend.position = "bottom",
          text = element_text(size = 16),
          axis.text.x = element_text(angle = 25, hjust = .9))


girafe(ggobj = ps.in.MN.ruca.plot) %>%
  girafe_options(opts_selection(type = "none"),
                 opts_sizing(rescale = FALSE))

Cross-tabs - RUCA

ps.in.MN.ct.ruca <- master.9 %>%
  filter(attended.ps == "Yes")

CrossTable(ps.in.MN.ct.ruca$Dem_Desc, ps.in.MN.ct.ruca$ps.in.MN, expected = TRUE, prop.t = FALSE, prop.c = FALSE, prop.chisq = FALSE)

 
   Cell Contents
|-------------------------|
|                       N |
|              Expected N |
|           N / Row Total |
|-------------------------|

 
Total Observations in Table:  88875 

 
                          | ps.in.MN.ct.ruca$ps.in.MN 
ps.in.MN.ct.ruca$Dem_Desc |                 In MN | Inside and outside MN |            Outside MN |             Row Total | 
--------------------------|-----------------------|-----------------------|-----------------------|-----------------------|
           Town/rural mix |                  9693 |                  2430 |                  3678 |                 15801 | 
                          |              9815.377 |              2465.934 |              3519.689 |                       | 
                          |                 0.613 |                 0.154 |                 0.233 |                 0.178 | 
--------------------------|-----------------------|-----------------------|-----------------------|-----------------------|
     Urban/town/rural mix |                 26232 |                  6433 |                  9159 |                 41824 | 
                          |             25980.528 |              6527.132 |              9316.340 |                       | 
                          |                 0.627 |                 0.154 |                 0.219 |                 0.471 | 
--------------------------|-----------------------|-----------------------|-----------------------|-----------------------|
           Entirely urban |                 19283 |                  5007 |                  6960 |                 31250 | 
                          |             19412.096 |              4876.934 |              6960.970 |                       | 
                          |                 0.617 |                 0.160 |                 0.223 |                 0.352 | 
--------------------------|-----------------------|-----------------------|-----------------------|-----------------------|
             Column Total |                 55208 |                 13870 |                 19797 |                 88875 | 
--------------------------|-----------------------|-----------------------|-----------------------|-----------------------|

 
Statistics for All Table Factors


Pearson's Chi-squared test 
------------------------------------------------------------
Chi^2 =  19.94633     d.f. =  4     p =  0.0005117295 


 

Next, lets check to see if the EDR of their high school is related to whether they attend a college inside or outside Minnesota.

The crosstabs below indicate that there is a relationship between the EDR category of a student’s high school and whether they attend a college inside or outside of Minnesota. The p-value was near zero.

EDR 9 has a signifcantly higher percentage of graduates attend college in Minnesota. EDR 9 had 67% of their graduates attend college in Minnesota while 60% of EDR 10 attended in Minnesota.

Chart - EDR

ps.in.MN.edr <- master.9 %>%
  filter(attended.ps == "Yes") %>%
  group_by(edr, ps.in.MN) %>%
  summarize(n = n()) %>%
  ungroup() %>%
  group_by(edr) %>%
  mutate(pct = n / sum(n)) %>%
  ungroup() %>%
  mutate(data_id = as.character(seq(n())))

ps.in.MN.edr.plot <- ggplot(ps.in.MN.edr, aes(ps.in.MN, pct, fill = edr, group = edr)) +
    geom_col_interactive(position = "dodge", aes(data_id = data_id, tooltip = paste(edr, "\nCollege in Minnesota or not: ", ps.in.MN, "\nNumber of students: ", comma(n, accuracy = 1), "\nPercent of students: ", percent(pct, accuracy = .1), sep = ""))) +
    geom_label(position = position_dodge(width = .9), aes(label = percent(pct, accuracy = .1)), show.legend = FALSE, color = "black", size = 5) +
    labs(x="", y = "", color="", title = "Percent of students attending college in\nMinnesota")+
    scale_y_continuous(labels=scales::percent)+
    theme_bar+
    scale_fill_manual(values= color.edr,
                       guide = guide_legend(ncol = 3)) +
    theme(legend.position = "bottom",
          text = element_text(size = 16),
          axis.text.x = element_text(angle = 25, hjust = .9))


girafe(ggobj = ps.in.MN.edr.plot) %>%
  girafe_options(opts_selection(type = "none"),
                 opts_sizing(rescale = FALSE))

Tabset - EDR

ps.in.MN.ct.edr <- master.9 %>%
  filter(attended.ps == "Yes")

CrossTable(ps.in.MN.ct.edr$edr, ps.in.MN.ct.edr$ps.in.MN, expected = TRUE, prop.t = FALSE, prop.c = FALSE, prop.chisq = FALSE)

 
   Cell Contents
|-------------------------|
|                       N |
|              Expected N |
|           N / Row Total |
|-------------------------|

 
Total Observations in Table:  88875 

 
                      | ps.in.MN.ct.edr$ps.in.MN 
  ps.in.MN.ct.edr$edr |                 In MN | Inside and outside MN |            Outside MN |             Row Total | 
----------------------|-----------------------|-----------------------|-----------------------|-----------------------|
   EDR 10 - Southeast |                 38120 |                 10198 |                 15052 |                 63370 | 
                      |             39364.624 |              9889.642 |             14115.734 |                       | 
                      |                 0.602 |                 0.161 |                 0.238 |                 0.713 | 
----------------------|-----------------------|-----------------------|-----------------------|-----------------------|
EDR 9 - South Central |                 17088 |                  3672 |                  4745 |                 25505 | 
                      |             15843.376 |              3980.358 |              5681.266 |                       | 
                      |                 0.670 |                 0.144 |                 0.186 |                 0.287 | 
----------------------|-----------------------|-----------------------|-----------------------|-----------------------|
         Column Total |                 55208 |                 13870 |                 19797 |                 88875 | 
----------------------|-----------------------|-----------------------|-----------------------|-----------------------|

 
Statistics for All Table Factors


Pearson's Chi-squared test 
------------------------------------------------------------
Chi^2 =  387.0265     d.f. =  2     p =  9.083746e-85 


 

What states are students going to college

The last piece here is to provide a map of the United States to see where students are going for college.

Outside of Minnesota, Southern MN graduates attend colleges in Wisconsin at the highest rate - 10.5%. This is followed by Iowa at 5.5%.

us_boundary <- read_sf('/Users/kellyasche/Library/CloudStorage/GoogleDrive-kasche@ruralmn.org/My Drive/Data Prep/R Projects/Shapefiles/US shapefiles/cb_2018_us_state_500k.shp', quiet = TRUE) %>%
  tigris::shift_geometry()
  
ps.state <- nsc.enrollment.location.join %>%
  select(PersonID, ps.statefp) %>%
  distinct(PersonID, ps.statefp) %>%
  group_by(ps.statefp) %>%
  summarize(n = n()) %>%
  ungroup() %>%
  mutate(pct = n / sum(n),
         pct.bins = cut(pct,
                        breaks = c(-1, 0, .01, .05, .10, 1),
                        labels = c("0", "0% - 1%", "1% - 5%", "5% - 10%", "More than 10%"))) %>% 
  right_join(us_boundary[,c(1,6,10)], by = c("ps.statefp" = "STATEFP")) %>%
  filter(!ps.statefp %in% c("60", "66", "69", "72", "78"))


ps.state.map <- ggplot(ps.state) +
  geom_sf_interactive(color = "grey85", aes(geometry = geometry, fill = pct.bins, data_id = NAME, tooltip = paste(NAME, "\nNumber of students enrolled in college from SW MN: ", comma(n, accuracy = 1), "\nPercent of students: ", percent(pct, accuracy = .1), sep = ""))) +
  coord_sf(crs = 'ESRI:102003') + 
  theme_sf+
  scale_fill_manual(values = brewer.pal(n = 4, "PuBu")) +
  labs(title = "Percent of students attending colleges in states") +
  theme(text = element_text(size = 18))

girafe(ggobj = ps.state.map, height_svg = 10, width_svg = 10) %>%
  girafe_options(opts_sizing(rescale = FALSE))

write_csv(master.9, "Data/SLEDS/Masters/Master-9.csv")